xDSD Gryphon: birth of a ‘head-fi’ legend
Oct 4, 2023 at 6:36 AM Post #3,691 of 3,842
Will the Gryphon have enough power to drive my HD800S?

Just in terms of power, in balanced mode, probably.

I used to have the ifi micro IDSD and it was enough to do the job, but I seem to recall reading that the gryphon has less power .
I dont need BT, so would I be better getting another micro IDSD?

The iDSD micro is a much more "high end" platform than the xDSD/xCAN which were "copy/pasta" merged into the "Gryphon".

For relative reference, the xDSD, xCAN and what eventually became the "Neo" were designed as replacements for the "nano" series at iFi, with a target pricing of ~ 200-300 USD retail (including the ~ 50 USD cost contribution of BT) in 2016..18. They use all across the board less quality passive components, headphone amplifiers are based on MAX97220 and Op-Amp's used are remarked OPA1679. Power supplies are pretty basic.

The iDSD micro was designed for a 399 USD retail price point in 2014 and upgraded to "black label" at 499 USD and now as "Signature/Finale" with minor changes mostly cosmetic (e.g.) sit's at 749 USD as a product essentially with an unchanged fundamental platform for nearly a decade.

As BL it uses in effect the highest quality passive components, Headphone Amp is TPA6120 and Op-Amp's are OPA1642 and power supplies are much more elaborate and sophisticated than "Gryphon".

So if sound quality is of main concern, I think the iDSD micro BL 2nd hand is an excellent choice.

Of the iFi products I designed I still have and use:

xDSD (portable/pocketable for travel with iEM's with iEMatch and Bose QC-25 only)

Zen Signature Stack (modified, upgraded power supplies) daily driver with planar headphones (also uses USB ipurifier and if USB cable)

iDSD micro Black Label - standby/backup for all headphones and used for a range of testing

Thor
 
Oct 4, 2023 at 9:42 AM Post #3,692 of 3,842
Just in terms of power, in balanced mode, probably.



The iDSD micro is a much more "high end" platform than the xDSD/xCAN which were "copy/pasta" merged into the "Gryphon".

For relative reference, the xDSD, xCAN and what eventually became the "Neo" were designed as replacements for the "nano" series at iFi, with a target pricing of ~ 200-300 USD retail (including the ~ 50 USD cost contribution of BT) in 2016..18. They use all across the board less quality passive components, headphone amplifiers are based on MAX97220 and Op-Amp's used are remarked OPA1679. Power supplies are pretty basic.

The iDSD micro was designed for a 399 USD retail price point in 2014 and upgraded to "black label" at 499 USD and now as "Signature/Finale" with minor changes mostly cosmetic (e.g.) sit's at 749 USD as a product essentially with an unchanged fundamental platform for nearly a decade.

As BL it uses in effect the highest quality passive components, Headphone Amp is TPA6120 and Op-Amp's are OPA1642 and power supplies are much more elaborate and sophisticated than "Gryphon".

So if sound quality is of main concern, I think the iDSD micro BL 2nd hand is an excellent choice.

Of the iFi products I designed I still have and use:

xDSD (portable/pocketable for travel with iEM's with iEMatch and Bose QC-25 only)

Zen Signature Stack (modified, upgraded power supplies) daily driver with planar headphones (also uses USB ipurifier and if USB cable)

iDSD micro Black Label - standby/backup for all headphones and used for a range of testing

Thor
Thanks for your very comprehensive reply. Thats very helpful.
 
Oct 4, 2023 at 11:20 PM Post #3,693 of 3,842
That would mean I would need a new cable for my HD800S but thanks for the reply

i am torn between the gryphon and the micro black label, which I used to own and stupidly let go :relaxed:

Don't know if it's still the case, but FWIW the Signature/Finale has been on sale recently for $400-ish (per the Deals thread)
 
Oct 15, 2023 at 5:02 PM Post #3,694 of 3,842
I dont think the Gryphon is hard to beat by a 1000+ daps.
For the last 2 weeks i was comparing the Gryphon with Fiio M15s and the Fiio beats the Gryphon in every department. Is more spacious, cleaner, more defined, better separated and even less forward in mids (which is surprising considering the fact that the Gryphon is rather laid back there).
Maybe when the Gryphon is connected to some higher end source, then it hits way above its price bracket but then its not a portable solution then.
Dont get me wrong, i love Gryphon's musical and muscular presentation but its technicalities arent as good as some 900-1000 dol daps.
But there are some other great things about it- a lot of power, XBass, XSpace, iem match, its light, well made, works like a charm and has IFI house sound.
Hey what would you recommend me, for using my Hifiman He1000V2? I mostly use it with my Mac, sometime for gaming on my computer or for watching music on my Apple TV. Would you recommend the Gryphon or is the M15S a noticeable improvement that’s worth considering?
 
Oct 17, 2023 at 2:33 AM Post #3,695 of 3,842
As I've increased my collection of sensitive IEMs the Gryphon has started to annoy me more so - the background noise is far too strong. I'm enjoying my Andromeda's more through my Qudelix nowadays which is dead silent. There definitely some shortcut taken, this level of hiss (even when using it as line out with speakers) should not be there... White noise galore.
Still hiss even when iematch is enabled?
 
Oct 17, 2023 at 3:01 AM Post #3,696 of 3,842
As I've increased my collection of sensitive IEMs the Gryphon has started to annoy me more so - the background noise is far too strong. I'm enjoying my Andromeda's more through my Qudelix nowadays which is dead silent. There definitely some shortcut taken, this level of hiss (even when using it as line out with speakers) should not be there... White noise galore.
Have the same issue with 3 IEM’s, all of them I have noise, I have to enable IEMatch with all of them, and they can say whatever they want, IEM does influence the sound, maybe not extreme, but enough to annoy me.

With overhead headphones I don’t have a issue, but I did not buy this device for overheads. I bought it as a high end portable for IEM.
 
Oct 17, 2023 at 3:19 AM Post #3,697 of 3,842
Have the same issue with 3 IEM’s, all of them I have noise, I have to enable IEMatch with all of them, and they can say whatever they want, IEM does influence the sound, maybe not extreme, but enough to annoy me.

Some notes, use the SE output, never balanced, for IEM's (or wire the 4.4mm connector for SE).

Use iEMatch only on the SE output. From the various notes and what I read here the 4.4mm iEMatch is not designed correctly, if we consider the 3.5mm iEMatch and iEMatch on the iDSD micro consider "correctly designed".

On the 3.5mm unit you get 12 dB level reduction with 2.5 Ohm source impedance to the iEM (this is mainly for single dynamic driver IEM's) and 24 dB level reduction with 1 Ohm source impedance to the iEM (this is mainly for multi driver BA and Hybrid IEM's).

For IEM's I think the original xDSD with a wired 3.5mm iEMatch will be a better choice. It's what I use anyway.

Thor
 
Oct 17, 2023 at 4:39 AM Post #3,698 of 3,842
Some notes, use the SE output, never balanced, for IEM's (or wire the 4.4mm connector for SE).

Use iEMatch only on the SE output. From the various notes and what I read here the 4.4mm iEMatch is not designed correctly, if we consider the 3.5mm iEMatch and iEMatch on the iDSD micro consider "correctly designed".

On the 3.5mm unit you get 12 dB level reduction with 2.5 Ohm source impedance to the iEM (this is mainly for single dynamic driver IEM's) and 24 dB level reduction with 1 Ohm source impedance to the iEM (this is mainly for multi driver BA and Hybrid IEM's).

For IEM's I think the original xDSD with a wired 3.5mm iEMatch will be a better choice. It's what I use anyway.

Thor
so what you are saying, getting a seperate IEmatch from ifi, is better than the build in? The problem I have with the 3,5. I don't know if I explain it correctly, to me it sound missing some stage, which I is there with the balanced. On but I have to use IEmatch, but both suck because of that. right now I am looking for a better solution than the xdsd gryphon (I love the all in one, but this way it is just annoying, and for my desktop stuff I already have a qutest and burson soloist 3x gt)
 
Last edited:
Oct 17, 2023 at 10:49 AM Post #3,699 of 3,842
so what you are saying, getting a seperate IEmatch from ifi, is better than the build in? The problem I have with the 3,5. I don't know if I explain it correctly, to me it sound missing some stage, which I is there with the balanced.

You may be "hearing things".

The balanced out is created by taking the output of SE Headphone amp and inverting it in a second copy of the original amp.

So, baring other factors the BAL output will have 9db more noise and 6 dB more odd order distortion.

In the end "iEMatch" is a resistor voltage divider based on a device from pro audio called the "power soak" which was used in studios to allow the guitar player use his amp and speakers in the studio, heavily overdriving the amp (and normally also making a lot of noise even when not playing and even more when) to keep the tone and lower noise and guitar levels to be usable.

So a correctly designed and implemented "iEMatch" circuit is essentially transparent and keeps a source impedance to the headphone that is suitable.

So first I would question why "to me it sound missing some stage, which I is there with the balanced." And if that is real, rather than a trick of perception.

If real, could it relate to cables?

Many variables.

On but I have to use IEmatch, but both suck because of that.

Remember, internally the balanced out is just the same as SE, with levels doubled over SE (+6dB) and extra noise and distortion. Strictly speaking on sound purity, SE on the Gryphon should always be better, if levels are sufficient.

If it is not and we exclude "you just like distortion" as cause ( I dislike it as excessively cynical, which is saying something), it would be worth digging a bit more.

right now I am looking for a better solution than the xdsd gryphon

For IEM I'd look at IEM specific amps. I would suggest the original xDSD for anything it drives well should be a better choice than Gryphon.

If the original xDSD is still too noisy (it is for example for Andromedas) just add a 3.5mm iEMatch (ultra for Andro's).

What I did for the iDSD micro which includes iEMatch was to make a headphone amplifier with around 120dB SNR at unity gain (aka Eco) and noise that scaled with gain (turbo has 24dB more gain but also 24dB more noise and distortion) and "soaking" (ultra has 24dB less signal but also 24dB less noise).

But is on the chunky side. Mind you, it is selectable between 12V, 6V, 2V, 0.5V and 0.125V for full signal and volume at clipping limit. All with pretty much even stevens noise.

As a result, it can drive headphones with a 48dB difference in sensitivity (say >124dB/V IEM's to 86dB/V AKG K-1000) in one device. Invariably being all things to all men means there will be some not served that well, according to their view, if only because "dedicated" obviously bests "multiple selection" every time, even if dedicated just removes the switched and hardwires them.

The "Gryphon" is a copy paste merge of products I designed, without actually understanding each original design and adding in "iEMatch" somehow, with the same single cheap switch used for the SE only iEMatch to handle both SE and BAL.

I cannot see how to do it witout excessive compromises.

The correct solution would have been to have one switch each for 3.5mm and 4.4mm and to offer both lo & hi settings for each (for balanced this needs a 4-pole 3 position switch and actually 18/30dB attenuation) and with 2.5 Ohm and 1 Ohm approximate output impedance.

Your perception of "3.5mm sounds worse" in the context sounds to me, based on long experience like one of the following:

Incorrect level matching - read the perceived difference are down to loudness as opposed to actual differences (the louder but otherwise identical source in A/B testing is preferred band described as being more dynamic and having better soundstage)

Badly designed 3.5mm cable that causes excessive crosstalk.

Expectation bias, we expected the much hyped and advertised balanced out to sound better than SE (when in reality the reverse is true) and that is what we hear, regardless of what the reality is.

Note, I am far from the "all sounds the same blind" position. But it is easy to deceive oneself. Or as Nietzsche put it charitably, we are all better impressionist artists than we credit ourselves.

Seeing the current approach to both engineering and commercial operations at iFi compared to what things were in my time and seeing today, I feel iFi has turned from the original motto of "we will not make something we would not buy retail with our own money" into pretty much the opposite.

But that may be my expectation bias. Maybe current iFi product are better than they ever were when I designed them and much better value for money than back then and I'm just prejudiced so I cannot see it.

Thor
 
Last edited:
Oct 17, 2023 at 9:36 PM Post #3,700 of 3,842
You may be "hearing things".

The balanced out is created by taking the output of SE Headphone amp and inverting it in a second copy of the original amp.

So, baring other factors the BAL output will have 9db more noise and 6 dB more odd order distortion.

In the end "iEMatch" is a resistor voltage divider based on a device from pro audio called the "power soak" which was used in studios to allow the guitar player use his amp and speakers in the studio, heavily overdriving the amp (and normally also making a lot of noise even when not playing and even more when) to keep the tone and lower noise and guitar levels to be usable.

So a correctly designed and implemented "iEMatch" circuit is essentially transparent and keeps a source impedance to the headphone that is suitable.

So first I would question why "to me it sound missing some stage, which I is there with the balanced." And if that is real, rather than a trick of perception.

If real, could it relate to cables?

Many variables.



Remember, internally the balanced out is just the same as SE, with levels doubled over SE (+6dB) and extra noise and distortion. Strictly speaking on sound purity, SE on the Gryphon should always be better, if levels are sufficient.

If it is not and we exclude "you just like distortion" as cause ( I dislike it as excessively cynical, which is saying something), it would be worth digging a bit more.



For IEM I'd look at IEM specific amps. I would suggest the original xDSD for anything it drives well should be a better choice than Gryphon.

If the original xDSD is still too noisy (it is for example for Andromedas) just add a 3.5mm iEMatch (ultra for Andro's).

What I did for the iDSD micro which includes iEMatch was to make a headphone amplifier with around 120dB SNR at unity gain (aka Eco) and noise that scaled with gain (turbo has 24dB more gain but also 24dB more noise and distortion) and "soaking" (ultra has 24dB less signal but also 24dB less noise).

But is on the chunky side. Mind you, it is selectable between 12V, 6V, 2V, 0.5V and 0.125V for full signal and volume at clipping limit. All with pretty much even stevens noise.

As a result, it can drive headphones with a 48dB difference in sensitivity (say >124dB/V IEM's to 86dB/V AKG K-1000) in one device. Invariably being all things to all men means there will be some not served that well, according to their view, if only because "dedicated" obviously bests "multiple selection" every time, even if dedicated just removes the switched and hardwires them.

The "Gryphon" is a copy paste merge of products I designed, without actually understanding each original design and adding in "iEMatch" somehow, with the same single cheap switch used for the SE only iEMatch to handle both SE and BAL.

I cannot see how to do it witout excessive compromises.

The correct solution would have been to have one switch each for 3.5mm and 4.4mm and to offer both lo & hi settings for each (for balanced this needs a 4-pole 3 position switch and actually 18/30dB attenuation) and with 2.5 Ohm and 1 Ohm approximate output impedance.

Your perception of "3.5mm sounds worse" in the context sounds to me, based on long experience like one of the following:

Incorrect level matching - read the perceived difference are down to loudness as opposed to actual differences (the louder but otherwise identical source in A/B testing is preferred band described as being more dynamic and having better soundstage)

Badly designed 3.5mm cable that causes excessive crosstalk.

Expectation bias, we expected the much hyped and advertised balanced out to sound better than SE (when in reality the reverse is true) and that is what we hear, regardless of what the reality is.

Note, I am far from the "all sounds the same blind" position. But it is easy to deceive oneself. Or as Nietzsche put it charitably, we are all better impressionist artists than we credit ourselves.

Seeing the current approach to both engineering and commercial operations at iFi compared to what things were in my time and seeing today, I feel iFi has turned from the original motto of "we will not make something we would not buy retail with our own money" into pretty much the opposite.

But that may be my expectation bias. Maybe current iFi product are better than they ever were when I designed them and much better value for money than back then and I'm just prejudiced so I cannot see it.

Thor
I thought that the Gryphon was truly balanced, so as your statement, the DAC inside Gryphon only has 2 channels?
 
Oct 18, 2023 at 7:45 AM Post #3,701 of 3,842
ie match does impact the sound in a negative way - it somehow compresses dynamics, it's obvious with good sensitive IEMs. I really don't want to see it in high end audiophile devices anymore - sure it can help with the noise, but I don't want to pay with sound quality for that.

I'm staying away from IFI for a while until they do some innovation and restructuring of their topology. It feels all their newest products are rehashes of old tech, and IEmatch needs to go for better silent circuits (preferably class A).

There is also no need for so much power, i'd much rather 250mw of silent muscular class A over hissy wimpy opamp based 500mw - it makes my headphones loud, great, but they sound strained and grainy.
 
Oct 19, 2023 at 11:41 AM Post #3,702 of 3,842
I thought that the Gryphon was truly balanced, so as your statement, the DAC inside Gryphon only has 2 channels?

The headphone Amplifier is single ended in, balanced out by inverting the output of the SE channel. This causes a 9dB SNR penalty over the SE output, when using the same headphone.

11856825.jpg


If you look at the PCB, U505 is the volume control, a relabelled NJW1159. It is a 2-channel volume control. For a balanced input / balanced output headphone amplifier that retains the balanced signal throughout you need two of these IC's.

The DAC itself uses the so-called "Birt" Circuit (named after David Birt from the BBC) that allows both SE and true BAL output from a balanced input. So like Zen DAC & iDSD Neo the fixed level line out is balanced.

But the headphone Amplifier is not balanced in -> out, only the output is balanced.

Thor
 
Oct 19, 2023 at 12:07 PM Post #3,703 of 3,842
The headphone Amplifier is single ended in, balanced out by inverting the output of the SE channel. This causes a 9dB SNR penalty over the SE output, when using the same headphone.

11856825.jpg


If you look at the PCB, U505 is the volume control, a relabelled NJW1159. It is a 2-channel volume control. For a balanced input / balanced output headphone amplifier that retains the balanced signal throughout you need two of these IC's.

The DAC itself uses the so-called "Birt" Circuit (named after David Birt from the BBC) that allows both SE and true BAL output from a balanced input. So like Zen DAC & iDSD Neo the fixed level line out is balanced.

But the headphone Amplifier is not balanced in -> out, only the output is balanced.

Thor

Interesting... Is this design typical for portable DAC/amps, or is iFi taking a shortcut here?
 
Oct 19, 2023 at 1:03 PM Post #3,705 of 3,842
Interesting... Is this design typical for portable DAC/amps, or is iFi taking a shortcut here?

I think a key issue is that the "Gryphon" is mainly a copy & paste merge of xDSD & xCAN I designed.

When originally designed the xDSD was meant as replacement for the 199 USD nano iDSD black label. As it had more inputs (SPDIF, Bluetooth) it was targeted at ~ 250 USD.

The xCAN started life as replacement of the 129 USD iCAN nano and the BT system was supposed to be optional, with perhaps ~ 150 USD target, on the request by S&M. It was to have a classic potentiometer as volume control, which pretty much rules out a true balanced circuit and as a fairly cheap product that's how it was designed.

It kind of jumped to the x series casework and other details because then S&M decided it would not possible to sell a balanced iCAN nano in the nano casework.

Also, the designs are quite old now, released in 2018 and designed quite a bit earlier. As in many ways the xDSD and xCAN ended up competing against each other, so it was clear that the next generation would merge the two devices conceptually. I left iFi before any real work was started on this.

Originally the idea was to get both the xDSD V2 AND a replacement for the iDSD micro using all the improvements (like volume control) and making them portable / transportable counterpart to dedicated "Desktop" units which, for the xDSD broadly came with the iDSD Neo and the parallel unit to iDSD micro (and signature, finale and diablo). What ifi offered me to do the job as external design studio was a joke, so I declined their offer. As a result there were no really new products in the micro category only warmed up leftovers and so far no desktop "micro" range.

The "Gryphon" appears "designed" by a copy/paste merge of xDSD & xCAN at the chinese factory, without taking advantage of the opportunity to go for a fully balanced circuit (the cost impact would have been minimal) and carrying a lot of circuitry in the audio path that could have been removed (so yes, there are more daisy chained active stages than really needed, which is never a good thing). It also did not get the dual loop headphone amplifier I tested out on the "Neo" based on what I see on the PCB, which would have lowered noise and distortion.

Lastly, yes, many other commercial "balanced" Headphone amplifiers actually turn the incoming signal to sigle ended and use the same system as I did in the xCAN and Diablo where the balanced out is created by inverting the SE Amplifier output, including some very pricey and "High End" ones. But yes, I consider a less than optimum approach to designing the audio path.

Thor
 

Users who are viewing this thread

Back
Top